Abstract: A research method based on Freeman code has been designed according to the importance of leaf teeth and leaf lobes in leaf morphology. Image pre-processing techniques are used to get the contours of the plant leaves. The corner detection algorithm based on chain code and the method of identifying the convexo-concave of peripherals based on chain code difference are combined to determine the feature points. An improved calculation method of target area is developed to calculate the areas of leaf teeth and leaf lobes. A leaf can be described by the following three aspects, the numbers of leaf teeth and leaf lobes, the proportions of their perimeters and areas to the entire perimeter and area. The experiments extract these three characteristics from 70 samples belonging to 3 species of plants. The characteristics are summarized by threshold range. Experimental results show that the accuracy of calculation areas has been greatly increased through the new method. These three features can reflect the different species of plant leaves and can be used as a new feature value to help identify and classify plant species.

Key words: image contours, leaf teeth, leaf lobes, area, feature extraction

摘要： 根据植物叶片形态特征中锯齿和叶裂的重要性，设计了基于链码的研究方法。通过对叶片图像预处理得到叶片的轮廓曲线、结合基于链码的拐点检测方法和基于链码差的边界凹凸性判别方法确定锯齿点和叶裂点、改进基于链码的目标面积计算方法计算锯齿和叶裂的面积。从锯齿和叶裂的面积和周长占整个叶片的面积和周长的比例以及叶片包含的锯齿和叶裂数三个方面对叶片进行描述。实验对3种共70张叶片提取特征，归纳特征阈值范围。实验结果表明新方法可以大大提高计算面积的精确度，并且这三个特征可以体现出不同种的植物叶片，可以作为新的特征值，为识别植物物种和分类提供帮助。

关键词: 图像轮廓, 锯齿, 叶裂, 面积, 特征提取